GroUSE: A Benchmark to Evaluate Evaluators in Grounded Question Answering
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912209787944960 |
|---|---|
| author | Muller, Sacha Loison, António Omrani, Bilel Viaud, Gautier |
| author_facet | Muller, Sacha Loison, António Omrani, Bilel Viaud, Gautier |
| contents | Retrieval-Augmented Generation (RAG) has emerged as a common paradigm to use Large Language Models (LLMs) alongside private and up-to-date knowledge bases. In this work, we address the challenges of using LLM-as-a-Judge when evaluating grounded answers generated by RAG systems. To assess the calibration and discrimination capabilities of judge models, we identify 7 generator failure modes and introduce GroUSE (Grounded QA Unitary Scoring of Evaluators), a meta-evaluation benchmark of 144 unit tests. This benchmark reveals that existing automated RAG evaluation frameworks often overlook important failure modes, even when using GPT-4 as a judge.
To improve on the current design of automated RAG evaluation frameworks, we propose a novel pipeline and find that while closed models perform well on GroUSE, state-of-the-art open-source judges do not generalize to our proposed criteria, despite strong correlation with GPT-4's judgement. Our findings suggest that correlation with GPT-4 is an incomplete proxy for the practical performance of judge models and should be supplemented with evaluations on unit tests for precise failure mode detection.
We further show that finetuning Llama-3 on GPT-4's reasoning traces significantly boosts its evaluation capabilities, improving upon both correlation with GPT-4's evaluations and calibration on reference situations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_06595 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | GroUSE: A Benchmark to Evaluate Evaluators in Grounded Question Answering Muller, Sacha Loison, António Omrani, Bilel Viaud, Gautier Computation and Language I.2.7 Retrieval-Augmented Generation (RAG) has emerged as a common paradigm to use Large Language Models (LLMs) alongside private and up-to-date knowledge bases. In this work, we address the challenges of using LLM-as-a-Judge when evaluating grounded answers generated by RAG systems. To assess the calibration and discrimination capabilities of judge models, we identify 7 generator failure modes and introduce GroUSE (Grounded QA Unitary Scoring of Evaluators), a meta-evaluation benchmark of 144 unit tests. This benchmark reveals that existing automated RAG evaluation frameworks often overlook important failure modes, even when using GPT-4 as a judge. To improve on the current design of automated RAG evaluation frameworks, we propose a novel pipeline and find that while closed models perform well on GroUSE, state-of-the-art open-source judges do not generalize to our proposed criteria, despite strong correlation with GPT-4's judgement. Our findings suggest that correlation with GPT-4 is an incomplete proxy for the practical performance of judge models and should be supplemented with evaluations on unit tests for precise failure mode detection. We further show that finetuning Llama-3 on GPT-4's reasoning traces significantly boosts its evaluation capabilities, improving upon both correlation with GPT-4's evaluations and calibration on reference situations. |
| title | GroUSE: A Benchmark to Evaluate Evaluators in Grounded Question Answering |
| topic | Computation and Language I.2.7 |
| url | https://arxiv.org/abs/2409.06595 |